|Publication number||US7944270 B2|
|Application number||US 11/918,583|
|Publication date||May 17, 2011|
|Filing date||Apr 7, 2006|
|Priority date||Apr 15, 2005|
|Also published as||DE102005017538A1, DE102005017538B4, DE502006003100D1, EP1875318A1, EP1875318B1, EP2051372A2, US20090219129, WO2006108575A1|
|Publication number||11918583, 918583, PCT/2006/3206, PCT/EP/2006/003206, PCT/EP/2006/03206, PCT/EP/6/003206, PCT/EP/6/03206, PCT/EP2006/003206, PCT/EP2006/03206, PCT/EP2006003206, PCT/EP200603206, PCT/EP6/003206, PCT/EP6/03206, PCT/EP6003206, PCT/EP603206, US 7944270 B2, US 7944270B2, US-B2-7944270, US7944270 B2, US7944270B2|
|Inventors||Urs Denier, Vivek Sharma|
|Original Assignee||Austriamicrosystems Ag|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (9), Non-Patent Citations (6), Referenced by (1), Classifications (15), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This is a U.S. national stage of International Application No. PCT/EP2006/03206, filed on Apr. 7, 2006.
This patent application claims the priority of German patent application No. 10 2005 017 538.4 filed Apr. 15, 2005, the disclosure content of which is hereby incorporated by reference.
The present invention relates to an arrangement for temperature compensation for a resistance and to a method for temperature compensation for a resistance.
Integrated resistances in semiconductor circuitry normally have a relatively high level of temperature dependency. However, it may be desirable to provide an integrated resistance in temperature-independent or temperature-compensated fashion. Such resistances which are insensitive to temperature are used, inter alia, in high-accuracy transconductance amplifiers, high-accuracy transimpedance amplifiers and in applications in medical engineering, for example in measuring equipment for blood sugar content.
Pages 248 to 250 in chapter 5.2 of the document by D. A. Johns, K. Martin: “Analog Integrated Circuit Design”, Toronto, Ontario, Canada, Wiley 1997 indicate how to provide on-chip resistances in thermally stable fashion. In this document, the transconductance of an operational amplifier is stabilized by virtue of there being a high-accuracy, thermally stable external resistance in the form of a non-integrated, discrete component.
The document by A. McLaren and K. Martin: “Generation of Accurate On-Chip Time Constants And Stable Transconductances”, IEEE Journal of Solid-State Circuits, Vol. 36, No. 4, April 2001 proposes a development of this principle such that an on-chip resistance is used instead of the external reference resistance. For the purpose of calibrating this resistance, complex analog circuitry is provided and also there is a need for a variable bias resistance split into numerous, parallel-connected elemental resistances, graduated in binary and thermometer code, see
With a temperature variation of around 60° Celsius, the transconductance scatter for this principle is nevertheless around 2.2%. Added to this is the fact that the convergence for such a circuit is relatively slow.
It is an object of the present invention to provide an arrangement and a method for temperature compensation for a resistance which are able to be implemented using integrated circuitry, have reduced circuit complexity and allow precise calibration.
This and other objects are attained in accordance with one aspect of the present invention directed to an arrangement for temperature compensation for a resistance:
Another aspect of the present invention is directed to a method for temperature compensation for a resistance having the following method steps:
The reference resistance can be provided by virtue of a switched capacitor forming the resistance being provided. The temperature compensation for a resistance is achieved by virtue of the resistance being controlled on the basis of a comparison result between the resistance value of the controllable resistance and the reference resistance, which the switched capacitor has as equivalent resistance value.
The switched-capacitor reference resistance is preferably of integrated design and has a temperature independency in practice. The temperature dependency of a switched capacitor is less than 100 ppm per Kelvin.
The controllable resistance can comprise a series circuit. The series circuit preferably has a fixed-value resistance and the controllable resistance.
The fixed-value resistance is preferably in the form of an integrated polyresistance. The controllable resistance is preferably in the form of a metal oxide semiconductor, MOS, transistor, with the resistance used being the controlled path and the control input used being the gate connection of the transistor.
In this case, the total resistance of the series circuit is kept constant in temperature-independent fashion by virtue of the total resistance of the series circuit being compared with the reference resistance, and the controllable resistance being controlled on the basis of any discrepancy, on the basis of the comparison result.
The controllable resistance can be controlled in a control loop such that the total resistance of the series circuit remains constant.
The reference resistance in the form of a switched capacitor preferably comprises a capacitance and at least one switch for periodic changeover. Such switched capacitors are also called switched capacitances.
In the case of a switched capacitance, the reference resistance is simulated by the switched capacitor. If a changeover switch connects the switched capacitance CS to an input voltage U then the capacitor receives a charge quantity Q which corresponds to the product of the capacitance value and the input voltage. In the other switch position, the capacitor outputs the same charge again. It therefore transfers this charge from the input to the output of the circuit in each switching period TS. This produces a flow of current I which, on average, sets itself to
The basic equivalence between the switched capacitance and a nonreactive resistance can therefore be indicated as
where R is the equivalent resistance. Notable is the linear relationship between the switching frequency fs and the equivalent admittance.
The capacitor in the switched-capacitor reference resistance may be in the form of what is known as a poly-poly or metal-metal structure, for example.
Consequently, the value of the reference resistance of a switched capacitor is dependent only on the switching frequency and the capacitance value. The temperature coefficient of the capacitance value is relatively low and is approximately 30 ppm per Kelvin, for example.
The practical implementation of such a circuit is relatively simple. It provides a quick and stable result, and requires only a few components to implement it. In addition, it affords a high level of flexibility in the choice of calibration parameters, which can be valuable for cost-efficient implementation.
The reference clock for actuating the reference resistance in the form of a switched capacitor is preferably provided by a reference clock generator. The reference clock generator preferably actuates the switch for periodic changeover.
In one embodiment of the invention, the controllable resistance is arranged in the return path of an amplifier. This means that it is possible to provide not only a resistance having high temperature constancy but also an amplifier having temperature-independent transconductance or transimpedance.
When the controllable resistance is in the form of a series circuit, the fixed-value resistance is preferably in the form of a polysilicon resistance in integrated form. The controllable resistance in this series circuit is a tunable metal oxide semiconductor, MOS, transistor which is preferably operated in the linear range.
In one embodiment of the invention, a common current source is provided which can be switchably connected either to a first or to a second current path. The first current path comprises the controllable resistance. The second current path comprises the reference resistance, which is in the form of a switched capacitor. The comparator has a respective input coupled to the first and to the second current path. Hence, both current paths carry a current of the same magnitude, which results in even higher precision for the temperature compensation for the controllable resistance.
As a further preference, the temperature compensation is produced by using a calibration current signal on the basis of the chopping principle. This chopping calibration signal can be produced using a small current source and a switch, for example. In this case, the clock rate is chosen such that the calibration signal's frequency is at a sufficient distance from the useful frequency range of a useful signal for processing. In this case, the controllable resistance is preferably arranged in a useful signal path. The useful signal path is set up for signal processing for a useful signal in a useful frequency range. The useful frequency range is different than the frequency of the reference clock. The chopped calibration signal is accordingly superimposed on the useful signal.
If, as envisaged as a preference, a current source is provided jointly for both current paths then process and temperature fluctuations cannot influence the performance of the temperature compensation, since the same current flows both through the controllable resistance and through the reference resistance.
The level of the calibration current on the basis of the reference clock is preferably low in order to avoid limiting the dynamic range of the useful signal path.
Preferably, the output signals from both current paths are demodulated with respect to the chopping signal and, after suitable filtering, are compared with one another. The filtering is preferably suitable for removing the useful signal, which has now been converted to the chopping frequency. In this case, the comparator is preferably a single-stage amplifier the output of which can be used to control the controllable resistance. This ensures that the resistance value of the controllable resistance follows the reference resistance accurately.
A calibration signal, which the useful signal contains on account of the chopping, outside of the useful frequency range with low amplitude can easily be removed. This can be done using analog filters or digital filters, for example. Digital filters can be used advantageously particularly when the useful signal is being digitized anyway for subsequent further processing.
In another embodiment of the invention, in addition to the controllable resistance a further controllable resistance having a controllable resistance value is provided and is arranged in a useful signal path. Accordingly the further controllable resistance with a controllable resistance value is used for the actual useful signal processing, while the controllable resistance to be compared with the reference resistance is formed in an auxiliary path. This controllable resistance can also be understood to be a dummy component. This dispenses with the useful signal being superimposed with a chopping signal. This dummy resistance is used to produce an output current which is compared with an equivalent reference current. The equivalent reference current is produced in a switched-capacitor resistance. The two currents are compared using an error amplifier. The error amplifier produces a control signal which is used to control the controllable resistance and the further controllable resistance. This control signal is used to alter the controllable resistance value until the two currents being compared by the comparator are the same. Preferably, the control signal is carried by a loop filter which guarantees the stability of the control loop. A temperature drift which nevertheless remains is dependent on the accuracy of a reference voltage, on the temperature drift in the capacitive reference element in the switched-capacitor reference resistance, on a temperature drift in a clock source and on a temperature drift in a preferably provided current mirror. To produce a stable reference voltage, a bandgap generator arrangement may be provided, for example. Such bandgaps have a temperature coefficient of less than 15 ppm per Kelvin, for example. Crystal clock generators have excellent, low temperature coefficients of very much lower than 1 ppm per Kelvin, for example. Accordingly, the reference capacitor in the switched-capacitor reference resistance makes the primary contribution to the temperature drift which still remains, for example at 43 ppm per Kelvin. Even with worst-case appraisal of all of these causes of a temperature drift, the latter remains significantly below 100 ppm per Kelvin overall, which is a significant improvement, with particularly simple circuit implementation in addition. The use of a copy of the controllable resistance as proposed in this development means that any errors which might arise as a result of charge injections from the switched-capacitor circuit into the useful signal path are avoided.
In this development too, the controllable resistance is preferably formed in a series circuit comprising a fixed-value resistance and the controllable resistance.
As a further preference, the error amplifier, namely the comparator, is designed using switched-capacitor technology.
A loop filter for filtering the control signal which is output by the comparator is also preferably designed using switched-capacitor technology.
The invention is explained in more detail below using a plurality of exemplary embodiments with reference to drawings, in which:
The reference resistance 2 in the form of a switched capacitor comprises a reference capacitor 5 and a switch 6. The switch 6 is actuated at a clock frequency fS of a reference clock. The equivalent resistance of this switched-capacitor reference resistance is calculated, on the basis of the specification, from the reciprocal product of the capacitance value CS of the capacitor 5 and the clock frequency fS of the reference clock actuating the switch 6.
The comparator compares the resistance value of the controllable resistance with the resistance value of the reference resistance. As soon as a temperature-related discrepancy in the resistance value of the controllable resistance 1 is found, the comparator outputs an error signal such that the resistance value of the controllable resistance 1 is tracked to the resistance value of the thermally stable reference resistance.
This achieves highly precise temperature compensation for the controllable resistance 1. Temperature dependencies in the resistance value of the controllable resistance 1 are corrected practically completely.
A common current source 11 is switchably connected either to the inverting input of the operational amplifier 7 or to a second current path 13 by means of a changeover switch 12. In this case, the common current source 11 outputs a calibration current ICAL. The non-inverting input of the operational amplifier 7 is connected to a reference potential connection. The second current path 13 comprises a reference resistance in the form of a switched capacitor 2, which in turn comprises a capacitance 5 and a changeover switch 6. The free connection of the capacitance is connected to a reference potential connection 14. Both the changeover switch 12 for the common current source 11 and the changeover switch 6 for the switched-capacitor reference resistance 2 are controlled by a reference clock at the reference frequency fS. The output of the operational amplifier 7 is connected to an input of a comparator 16 via a bandpass filter 15 to form a first current path. Similarly, the second current path 13, which comprises the reference resistance 2, is connected to a second input of the comparator 16 via the bandpass filter 15. The output of the comparator 16 is connected to the gate connection of the controllable resistance 1′ in the form of a transistor, to form a control loop.
Temperature fluctuations in the fixed-value resistance 4 in the form of an integrated polyresistance are automatically compensated for by using a calibration. The MOS transistor 1′ is operated in its linear range. The total resistance of this series circuit 4, 1′ is compared with the reference resistance 2, the equivalent resistance of the switched capacitor 5.
The reference capacitor 5 is in the form of a poly-poly or metal-metal structure. The equivalent reference resistance of the reference resistance 2 is dependent only on the switching frequency fS and the capacitance value of the capacitor 5, whose temperature coefficient is relatively low, for example approximately 30 ppm per Kelvin.
A small calibration current is produced by means of the calibration current source 11 using commutation. This calibration current is used to measure the resistance 4, 1′ and also the reference resistance 2 and to compare them. To this end, periodic changeover at the switching frequency fS is provided. A filtering unit 15 and a comparison unit 16 are used to produce a control signal which controls the gate-source voltage of the MOS transistor 1′ and hence changes the resistance of the latter's controlled path.
The practical implementation is very simple and the calibration can take place continuously in the background without interrupting the useful signal processing between the generator 8 and the output 10 of the useful signal path. This works very quickly and requires only a few components. Furthermore, a high level of flexibility is provided.
In the present case, the calibration is accordingly performed using a chopped calibration current signal. This chopping signal is easily produced with the calibration current source 11 and the switch 12. The clock rate fs for the switch 12 is chosen such that the frequency of the calibration signal is sufficiently separated from the useful frequency range, supplied by the useful signal source 8, the transimpedance amplifier 7 operating in this useful frequency range. The expression “sufficiently separated” refers to a magnitude such as to enable removal of the calibration frequency from the useful signal. Process and temperature fluctuations in the calibration current from the source 11 have no influence on the properties and quality of the calibration operation, since the same calibration current ICAL flows both through the reference resistance in the second current path 13 and through the first current path, which comprises the fixed-value resistance 4 and the controlled resistance 1′. The magnitude of the calibration current may be small in order to avoid limiting the dynamic range of the useful signal processing.
The bandpass filter 15 comprises means for demodulating the chopping signal using a simple chopper and downstream filtering. The filtering is used to eliminate the useful signal, which has now been converted to the chopping frequency. The comparison in the comparator 16 can be performed using a simple single-stage amplifier whose output is used to tune the resistance 1′. This ensures that the target resistance 1′, 4 follows the reference resistance 2 as a series circuit.
In relation to the useful signal at the output of the amplifier 7, the calibration signal is small and is outside of the useful frequency range, it can easily be removed using the low pass filter 9. Even more simple is the removal of the calibration components in the useful signal in the digital domain after analog/digital conversion.
In the present case, the calibrating principle is shown by means of a transimpedance amplifier 7. It goes without saying that the principle can also be applied, within the context of the invention, to other circuit configurations with modifications which are within the scope of technical action. By way of example, the calibration signal may be a voltage or a current, and the type of comparison and of calibration may be current comparison, voltage comparison or similar. The level and frequency of the calibration signal may accordingly match the application. In this case, it is desirable to keep the frequency of the calibration signal sufficiently separated from the useful signal frequency band, be it at a higher or at a lower frequency. In this way it is possible to use a very small calibration signal in terms of its amplitude. A simplified filter design is possible. The filters may be RC, GmC or SC filters.
The further controllable resistance 21 comprises a metal oxide semiconductor, MOS, transistor with a control input and a controlled path. The control voltage for the controllable resistance 21 is provided by a circuit as shown in
The circuit explained below which is shown in
The comparator 27 is itself in the form of a switched-capacitor amplifier. For this, a return path comprises three parallel-connected paths. A first path between the output of the comparator 27 and its inverting input comprises a switch 35 which can be activated in a first clock phase φ1. A second path comprises a capacitor 36 which is connected in series with a switch 37 for operation in a clock phase φ2. A second capacitor 38 likewise forms a series circuit with a switch 39 in a third return path. The switch 39 is closed in the second to fourth clock phases φ2, φ3, φ4. Further switches 40, 41 are used to connect a node between the first and second capacitors 36, 38 and the associated switch 37, 39 to a reference potential connection 42.
In this case, the further switch 40, which is associated with the first capacitor 36, is closed in the first and third clock phases φ1, φ3. The further switch 41, which is associated with the second capacitor 38, is closed in the first clock phase φ1. The output of the comparator 27, which implements an error amplifier, is connected to the output 22 via a loop filter. The loop filter comprises a switched capacitor 43 which is connected as switched capacitor between the output of the amplifier 27 and an inverting input of a further amplifier 44. Four switches on the switched capacitor 43 are used to charge it in the third clock phase φ3 with the output signal from the amplifier 27 and to discharge it in the fourth clock phase φ4 to the input of the further amplifier 44. The further amplifier 44 has a capacitance 45 in the return path between its output, which is connected to the output 22, and its inverting input. The non-inverting input of the amplifier 44 is connected to ground potential. The capacitors 43, 45 are used to stabilize the control loop. The capacitors 36, 38 in the return path of the error amplifier define the residual error.
In line with
In detail, the circuit's capacitors 30, 36, 38 in the reference resistance and in the return path of the error amplifier are reset in the first clock phase φ1. The output of the operational amplifier 27 is set to the analog ground voltage VAGND. At the output of the amplifier 44, the control voltage remains unchanged.
In the second clock phase φ2, the capacitors 36, 38 in the return path of the error amplifier are connected in parallel. At the start of this clock phase, the reference resistance with the capacitor 30 is charged to the analog reference voltage VAGND, and this charge is transferred to the two capacitors 36, 38. During the second clock phase, the capacitors 36, 38 are discharged by the output current IOUT in an integration cycle. At the end of the second clock phase, the charge difference between the reference voltage VAGND multiplied by the reference capacitor 30 and the output current Iout multiplied by the clock period of the second phase φ2 is stored on the capacitors 36, 38.
In the third clock phase φ3, the charge difference described is amplified by transferring the charge from the capacitor 36 to the capacitor 38. For this, capacitor 36 is shorted to the analog reference voltage VAGND. The output of the error amplifier 27 is sampled to the switched capacitor 43.
In the fourth and final clock phase φ4, the charge is transferred from the capacitor 43 to the capacitor 45 of the loop filter and controls the gate of the transistor 21′ and of the transistor 21 in
The scope of protection of the invention is not limited to the examples given hereinabove. The invention is embodied in each novel characteristic and each combination of characteristics, which includes every combination of any features which are stated in the claims, even if this feature or combination of features is not explicitly stated in the examples.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3231827 *||Jan 22, 1963||Jan 25, 1966||Fernseh Gmbh||Variable gain transistor amplifier|
|US4781195||Dec 2, 1987||Nov 1, 1988||The Boc Group, Inc.||Blood monitoring apparatus and methods with amplifier input dark current correction|
|US5124593 *||Sep 26, 1990||Jun 23, 1992||National Semiconductor Corporation||Continuous-time filter tuning circuit and method|
|US5912589||Jun 26, 1997||Jun 15, 1999||Lucent Technologies||Arrangement for stabilizing the gain bandwidth product|
|US5973524||Mar 25, 1998||Oct 26, 1999||Silsym, Inc.||Obtaining accurate on-chip time-constants and conductances|
|US6646539||Oct 31, 2001||Nov 11, 2003||Infineon Technologies Ag||Temperature-compensated semiconductor resistor and semiconductor integrated circuit having the semiconductor resistor|
|DE69626912T2||Sep 3, 1996||Oct 2, 2003||Harris Corp||Aktive Vorspannungssteuerung für zeitkontinuierliche integrierte Schaltung|
|EP0477694A2||Sep 13, 1991||Apr 1, 1992||National Semiconductor Corporation||Continuous-time filter tuning circuit and method|
|EP0766385A2||Sep 3, 1996||Apr 2, 1997||Harris Corporation||Sample data-biasing of continuous time integrated circuit|
|1||A. Maclaren, et al., "Generation of Accurate On-Chip Time Constants and Stable Transductances", IEEE Journal of Solid-State Circuits, vol. 36, No. 4, Apr. 2001, pp. 691-695.|
|2||D.A. Johns, et al., "Analog Integrated Circuit Design", Toronto, Canada, Wiley 1997, Chapter 5, pp. 248-250.|
|3||D.M. Pietruszynski, et al., "A 50-Mbit/s CMOS Monolithic Optical Receiver", IEEE Journal of Solid-State Circuits, vol. 23, No. 6, Dec. 1988, pp. 1426-1433.|
|4||J.P. Hein, et al., "z-Domain Model for Discret-Time PLL's", IEEE Trans. Circuits and Systems, vol. 35, Nov. 1988, pp. 1393-1400.|
|5||K.N. Leung, et al., "A Sub-1-V 15-ppm/° C. CMOS Bandgap Voltage Reference without Requiring Low Threshold Voltage Device", IEEE Journal of Solid-State Circuits, vol. 37, No. 4, Apr. 2002, pp. 526-530.|
|6||Vanalog, "Tuning Fiork Quartz Crystal", Datasheet WX26B, 2006.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8922266||Feb 13, 2013||Dec 30, 2014||Megachips Corporation||Calibration circuit|
|International Classification||H01L35/00, H03K17/78, H03K3/42, H01L37/00|
|Cooperative Classification||H03F3/087, H03F3/45475, H03F2200/447, H03F3/45928, H03F1/08, H03F2203/45524|
|European Classification||H03F1/08, H03F3/45S1K, H03F3/08I, H03F3/45S3K|
|Feb 4, 2009||AS||Assignment|
Owner name: AUSTRIAMICROSYSTEMS AG, AUSTRIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:DENIER, URS;SHARMA, VIVEK;REEL/FRAME:022205/0102;SIGNINGDATES FROM 20090115 TO 20090116
Owner name: AUSTRIAMICROSYSTEMS AG, AUSTRIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:DENIER, URS;SHARMA, VIVEK;SIGNING DATES FROM 20090115 TO20090116;REEL/FRAME:022205/0102
|Mar 3, 2009||AS||Assignment|
Owner name: AUSTRIAMICROSYSTEMS AG, AUSTRIA
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE ASSIGNEE S COUNTRY CODE FROM CH TO AT; AND TO CORRECT THE SPELLING OF ARRNAGEMENT TO ARRANGEMENT IN THE TITLE PREVIOUSLY RECORDED ON REEL 022205 FRAME 0102;ASSIGNORS:DENIER, URS;SHARMA, VIVEK;REEL/FRAME:022335/0660;SIGNING DATES FROM 20090115 TO 20090116
Owner name: AUSTRIAMICROSYSTEMS AG, AUSTRIA
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE ASSIGNEE S COUNTRY CODE FROM CH TO AT; AND TO CORRECT THE SPELLING OF ARRNAGEMENT TO ARRANGEMENT IN THE TITLE PREVIOUSLY RECORDED ON REEL 022205 FRAME 0102. ASSIGNOR(S) HEREBY CONFIRMS THE ASSIGNMENT;ASSIGNORS:DENIER, URS;SHARMA, VIVEK;SIGNING DATES FROM 20090115 TO 20090116;REEL/FRAME:022335/0660
|Nov 12, 2014||FPAY||Fee payment|
Year of fee payment: 4